[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-Id: <FC2C38E0-D25F-4FB3-B7AA-E7F973EA5B69@net-swift.com>
Date: Thu, 26 Jun 2025 16:57:40 +0800
From: "mengyuanlou@...-swift.com" <mengyuanlou@...-swift.com>
To: Andrew Lunn <andrew@...n.ch>
Cc: netdev@...r.kernel.org,
michal.swiatkowski@...ux.intel.com,
kuba@...nel.org,
pabeni@...hat.com,
horms@...nel.org,
andrew+netdev@...n.ch,
duanqiangwen@...-swift.com,
linglingzhang@...stnetic.com,
jiawenwu@...-swift.com
Subject: Re: [PATCH net-next v2 08/12] net: txgbevf: add phylink check flow
> 2025年6月26日 15:55,Andrew Lunn <andrew@...n.ch> 写道:
>
> On Thu, Jun 26, 2025 at 02:31:36PM +0800, mengyuanlou@...-swift.com wrote:
>>
>>
>>> 2025年6月26日 01:08,Andrew Lunn <andrew@...n.ch> 写道:
>>>
>>> On Wed, Jun 25, 2025 at 06:20:54PM +0800, Mengyuan Lou wrote:
>>>> Add phylink support to wangxun 10/25/40G virtual functions.
>>>
>>> What do you gain by having a phylink instance for a VF?
>>>
>>> All the ops you define are basically NOPs. What is phylink actually
>>> doing?
>>
>> I think phylink helps me monitor the changes in link status and
>> automatically set the netif status.
>> Besides, I can also directly use the get_link_ksettings interface.
>
> But does a VF have anything to return for get_link_ksettings?
Just link status and phy mode set by myself.
>
> My understanding of a VF is that it connects to a port of an embedded
> switch. There is no media as such for a VF. The media itself is
> connected to another port of the switch, and you can get the link
> status via the PF. Also, even if the media is down, you should be
> still able to do VF to VF, via the switch.
>
> What do other drivers do which implement VFs? i40? mlx5?
>
None of the vf drivers do it in this way.
Ok, I will change it.
> Andrew
>
Powered by blists - more mailing lists